NB: b3 is currently experimental, the API may change without warning.
b3 helps you compare two Git revisions, typically main and a branch, to understand how your changes affect performance. It is designed for CI but also works during local development. b3 runs the revisions in pairs and uses a Bayesian bootstrap with a small regression to estimate the branch's effect, account for drift and run order, and report the uncertainty around the results.
b3 --baseline main --candidate HEAD --repetitions 30 --output-dir benchmark/ -- Rscript benchmark.RRun b3 --help for the full set of options.
Each run writes to --output-dir:
config.json: run's parameters and resolved revisions.benchmark.log: one JSON line per individual run.measurements.csv: paired baseline/candidate timings.posterior.csv: Bayesian bootstrap draws.report.txt: human-readable summary.
Dual licensed under either of Apache License, Version 2.0 or MIT license at your option.